src/backup/chunk_store.rs

   1 use failure::*;
   2 use std::path::{Path, PathBuf};
   3 use std::io::Write;
   4 use std::time::Duration;
   5
   6 use openssl::sha;
   7 use std::sync::Mutex;
   8
   9 use std::fs::File;
  10 use std::os::unix::io::AsRawFd;
  11
  12 use crate::tools;
  13
  14 pub struct GarbageCollectionStatus {
  15     pub used_bytes: usize,
  16     pub used_chunks: usize,
  17     pub disk_bytes: usize,
  18     pub disk_chunks: usize,
  19 }
  20
  21 impl Default for GarbageCollectionStatus {
  22     fn default() -> Self {
  23         GarbageCollectionStatus {
  24             used_bytes: 0,
  25             used_chunks: 0,
  26             disk_bytes: 0,
  27             disk_chunks: 0,
  28         }
  29     }
  30 }
  31
  32 /// File system based chunk store
  33 pub struct ChunkStore {
  34     name: String, // used for error reporting
  35     pub (crate) base: PathBuf,
  36     chunk_dir: PathBuf,
  37     mutex: Mutex<bool>,
  38     _lockfile: File,
  39 }
  40
  41 // TODO: what about sysctl setting vm.vfs_cache_pressure (0 - 100) ?
  42
  43 fn digest_to_prefix(digest: &[u8]) -> PathBuf {
  44
  45     let mut buf = Vec::<u8>::with_capacity(2+1+2+1);
  46
  47     const HEX_CHARS: &'static [u8; 16] = b"0123456789abcdef";
  48
  49     buf.push(HEX_CHARS[(digest[0] as usize) >> 4]);
  50     buf.push(HEX_CHARS[(digest[0] as usize) &0xf]);
  51     buf.push(HEX_CHARS[(digest[1] as usize) >> 4]);
  52     buf.push(HEX_CHARS[(digest[1] as usize) & 0xf]);
  53     buf.push('/' as u8);
  54
  55     let path = unsafe { String::from_utf8_unchecked(buf)};
  56
  57     path.into()
  58 }
  59
  60 impl ChunkStore {
  61
  62     fn chunk_dir<P: AsRef<Path>>(path: P) -> PathBuf {
  63
  64         let mut chunk_dir: PathBuf = PathBuf::from(path.as_ref());
  65         chunk_dir.push(".chunks");
  66
  67         chunk_dir
  68     }
  69
  70     pub fn create<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
  71
  72         let base: PathBuf = path.into();
  73
  74         if !base.is_absolute() {
  75             bail!("expected absolute path - got {:?}", base);
  76         }
  77
  78         let chunk_dir = Self::chunk_dir(&base);
  79
  80         if let Err(err) = std::fs::create_dir(&base) {
  81             bail!("unable to create chunk store '{}' at {:?} - {}", name, base, err);
  82         }
  83
  84         if let Err(err) = std::fs::create_dir(&chunk_dir) {
  85             bail!("unable to create chunk store '{}' subdir {:?} - {}", name, chunk_dir, err);
  86         }
  87
  88         // create 64*1024 subdirs
  89         let mut last_percentage = 0;
  90
  91         for i in 0..64*1024 {
  92             let mut l1path = chunk_dir.clone();
  93             l1path.push(format!("{:04x}", i));
  94             if let Err(err) = std::fs::create_dir(&l1path) {
  95                 bail!("unable to create chunk store '{}' subdir {:?} - {}", name, l1path, err);
  96             }
  97             let percentage = (i*100)/(64*1024);
  98             if percentage != last_percentage {
  99                 eprintln!("Percentage done: {}", percentage);
 100                 last_percentage = percentage;
 101             }
 102         }
 103
 104         Self::open(name, base)
 105     }
 106
 107     pub fn open<P: Into<PathBuf>>(name: &str, path: P) -> Result<Self, Error> {
 108
 109         let base: PathBuf = path.into();
 110
 111         if !base.is_absolute() {
 112             bail!("expected absolute path - got {:?}", base);
 113         }
 114
 115         let chunk_dir = Self::chunk_dir(&base);
 116
 117         if let Err(err) = std::fs::metadata(&chunk_dir) {
 118             bail!("unable to open chunk store '{}' at {:?} - {}", name, chunk_dir, err);
 119         }
 120
 121         let mut lockfile_path = base.clone();
 122         lockfile_path.push(".lock");
 123
 124         // make sure only one process/thread/task can use it
 125         let lockfile = tools::open_file_locked(
 126             lockfile_path, Duration::from_secs(10))?;
 127
 128         Ok(ChunkStore {
 129             name: name.to_owned(),
 130             base,
 131             chunk_dir,
 132             _lockfile: lockfile,
 133             mutex: Mutex::new(false)
 134         })
 135     }
 136
 137     pub fn touch_chunk(&self, digest:&[u8]) -> Result<(), Error> {
 138
 139         let mut chunk_path = self.chunk_dir.clone();
 140         let prefix = digest_to_prefix(&digest);
 141         chunk_path.push(&prefix);
 142         let digest_str = tools::digest_to_hex(&digest);
 143         chunk_path.push(&digest_str);
 144
 145         const UTIME_NOW: i64 = ((1 << 30) - 1);
 146         const UTIME_OMIT: i64 = ((1 << 30) - 2);
 147
 148         let times: [libc::timespec; 2] = [
 149             libc::timespec { tv_sec: 0, tv_nsec: UTIME_NOW },
 150             libc::timespec { tv_sec: 0, tv_nsec: UTIME_OMIT }
 151         ];
 152
 153         use nix::NixPath;
 154
 155         let res = chunk_path.with_nix_path(|cstr| unsafe {
 156             libc::utimensat(-1, cstr.as_ptr(), &times[0], libc::AT_SYMLINK_NOFOLLOW)
 157         })?;
 158
 159         if let Err(err) = nix::errno::Errno::result(res) {
 160             bail!("updata atime failed for chunk {:?} - {}", chunk_path, err);
 161         }
 162
 163         Ok(())
 164     }
 165
 166     pub fn read_chunk(&self, digest:&[u8], buffer: &mut Vec<u8>) -> Result<(), Error> {
 167
 168         let mut chunk_path = self.chunk_dir.clone();
 169         let prefix = digest_to_prefix(&digest);
 170         chunk_path.push(&prefix);
 171         let digest_str = tools::digest_to_hex(&digest);
 172         chunk_path.push(&digest_str);
 173
 174         let mut f = std::fs::File::open(&chunk_path)?;
 175
 176         let stat = nix::sys::stat::fstat(f.as_raw_fd())?;
 177         let size = stat.st_size as usize;
 178
 179         if buffer.capacity() < size {
 180             let mut newsize =  buffer.capacity();
 181             while newsize < size { newsize = newsize << 1; }
 182             let additional = newsize - buffer.len();
 183             buffer.reserve_exact(additional);
 184         }
 185         unsafe { buffer.set_len(size); }
 186
 187         use std::io::Read;
 188
 189         f.read_exact(buffer.as_mut_slice())?;
 190
 191         Ok(())
 192     }
 193
 194     pub fn get_chunk_iterator(
 195         &self,
 196     ) -> Result<
 197         impl Iterator<Item = Result<tools::fs::ReadDirEntry, Error>>,
 198         Error
 199     > {
 200         use nix::dir::Dir;
 201         use nix::fcntl::OFlag;
 202         use nix::sys::stat::Mode;
 203
 204         let base_handle = match Dir::open(
 205             &self.chunk_dir, OFlag::O_RDONLY, Mode::empty()) {
 206             Ok(h) => h,
 207             Err(err) => bail!("unable to open store '{}' chunk dir {:?} - {}",
 208                               self.name, self.chunk_dir, err),
 209         };
 210
 211         let mut verbose = true;
 212         let mut last_percentage = 0;
 213
 214         Ok((0..0x10000).filter_map(move |index| {
 215             let percentage = (index * 100) / 0x10000;
 216             if last_percentage != percentage {
 217                 last_percentage = percentage;
 218                 eprintln!("percentage done: {}", percentage);
 219             }
 220             let subdir: &str = &format!("{:04x}", index);
 221             match tools::fs::read_subdir(base_handle.as_raw_fd(), subdir) {
 222                 Err(e) => {
 223                     if verbose {
 224                         eprintln!("Error iterating through chunks: {}", e);
 225                         verbose = false;
 226                     }
 227                     None
 228                 }
 229                 Ok(iter) => Some(iter),
 230             }
 231         })
 232         .flatten()
 233         .filter(|entry| {
 234             // Check that the file name is actually a hash! (64 hex digits)
 235             let entry = match entry {
 236                 Err(_) => return true, // pass errors onwards
 237                 Ok(ref entry) => entry,
 238             };
 239             let bytes = entry.file_name().to_bytes();
 240             if bytes.len() != 64 {
 241                 return false;
 242             }
 243             for b in bytes {
 244                 if !b.is_ascii_hexdigit() {
 245                     return false;
 246                 }
 247             }
 248             true
 249         }))
 250     }
 251
 252     pub fn sweep_unused_chunks(&self, status: &mut GarbageCollectionStatus) -> Result<(), Error> {
 253         use nix::sys::stat::fstatat;
 254
 255         let now = unsafe { libc::time(std::ptr::null_mut()) };
 256
 257         for entry in self.get_chunk_iterator()? {
 258             let (dirfd, entry) = match entry {
 259                 Ok(entry) => (entry.parent_fd(), entry),
 260                 Err(_) => continue, // ignore errors
 261             };
 262
 263             let file_type = match entry.file_type() {
 264                 Some(file_type) => file_type,
 265                 None => bail!("unsupported file system type on chunk store '{}'", self.name),
 266             };
 267             if file_type != nix::dir::Type::File {
 268                 continue;
 269             }
 270
 271             let filename = entry.file_name();
 272             if let Ok(stat) = fstatat(dirfd, filename, nix::fcntl::AtFlags::AT_SYMLINK_NOFOLLOW) {
 273                 let age = now - stat.st_atime;
 274                 //println!("FOUND {}  {:?}", age/(3600*24), filename);
 275                 if age/(3600*24) >= 2 {
 276                     println!("UNLINK {}  {:?}", age/(3600*24), filename);
 277                     let res = unsafe { libc::unlinkat(dirfd, filename.as_ptr(), 0) };
 278                     if res != 0 {
 279                         let err = nix::Error::last();
 280                         bail!(
 281                             "unlink chunk {:?} failed on store '{}' - {}",
 282                             filename,
 283                             self.name,
 284                             err,
 285                         );
 286                     }
 287                 } else {
 288                     status.disk_chunks += 1;
 289                     status.disk_bytes += stat.st_size as usize;
 290
 291                 }
 292             }
 293         }
 294         Ok(())
 295     }
 296
 297     pub fn insert_chunk(&self, chunk: &[u8]) -> Result<(bool, [u8; 32]), Error> {
 298
 299         // fixme: use Sha512/256 when available
 300         let mut hasher = sha::Sha256::new();
 301         hasher.update(chunk);
 302
 303         let digest = hasher.finish();
 304
 305         //println!("DIGEST {}", tools::digest_to_hex(&digest));
 306
 307         let mut chunk_path = self.chunk_dir.clone();
 308         let prefix = digest_to_prefix(&digest);
 309         chunk_path.push(&prefix);
 310         let digest_str = tools::digest_to_hex(&digest);
 311         chunk_path.push(&digest_str);
 312
 313         let lock = self.mutex.lock();
 314
 315         if let Ok(metadata) = std::fs::metadata(&chunk_path) {
 316             if metadata.is_file() {
 317                  return Ok((true, digest));
 318             } else {
 319                 bail!("Got unexpected file type on store '{}' for chunk {}", self.name, digest_str);
 320             }
 321         }
 322
 323         let mut tmp_path = chunk_path.clone();
 324         tmp_path.set_extension("tmp");
 325         let mut f = std::fs::File::create(&tmp_path)?;
 326         f.write_all(chunk)?;
 327
 328         if let Err(err) = std::fs::rename(&tmp_path, &chunk_path) {
 329             if let Err(_) = std::fs::remove_file(&tmp_path)  { /* ignore */ }
 330             bail!(
 331                 "Atomic rename on store '{}' failed for chunk {} - {}",
 332                 self.name,
 333                 digest_str,
 334                 err,
 335             );
 336         }
 337
 338         //println!("PATH {:?}", chunk_path);
 339
 340         drop(lock);
 341
 342         Ok((false, digest))
 343     }
 344
 345     pub fn relative_path(&self, path: &Path) -> PathBuf {
 346
 347         let mut full_path = self.base.clone();
 348         full_path.push(path);
 349         full_path
 350     }
 351
 352     pub fn base_path(&self) -> PathBuf {
 353         self.base.clone()
 354     }
 355 }
 356
 357
 358 #[test]
 359 fn test_chunk_store1() {
 360
 361     let mut path = std::fs::canonicalize(".").unwrap(); // we need absulute path
 362     path.push(".testdir");
 363
 364     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 365
 366     let chunk_store = ChunkStore::open("test", &path);
 367     assert!(chunk_store.is_err());
 368
 369     let chunk_store = ChunkStore::create("test", &path).unwrap();
 370     let (exists, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 371     assert!(!exists);
 372
 373     let (exists, _) = chunk_store.insert_chunk(&[0u8, 1u8]).unwrap();
 374     assert!(exists);
 375
 376
 377     let chunk_store = ChunkStore::create("test", &path);
 378     assert!(chunk_store.is_err());
 379
 380     if let Err(_e) = std::fs::remove_dir_all(".testdir") { /* ignore */ }
 381 }